We had experienced something like this a few times. In most cases was enough to shutdown Jackrabbit, delete the contents of the repository/workspaces/<name>/index directory (for each workspace if you more than one) and start Jackrabbit again. It will re-index all the contents, and it should be faster and less troublesome than exporting/importing again. Depending on how much data you have loaded, the process could take from a few minutes to a few hours.
Hope it helps. Alessandro On Tue, Apr 1, 2008 at 7:41 AM, Juerg Meier <[EMAIL PROTECTED]> wrote: > Just for completeness... > > I had to reload the JCR from ground-up (export - remove repository - > import), now it works again. And I really hope this won't happen in > production... > > -- Juerg > > ________________________________________ > From: Juerg Meier [EMAIL PROTECTED] > Sent: Monday, March 31, 2008 3:19 PM > To: [email protected] > Subject: XPATH problem > > Hi, > > I have a problem in category "query should return something, but it > doesn't", too. > > One subnode of the rep does not seem to be reachable anymore by xpath > queries. The problem exists since I shut down my app, and temporarely > startet magnolia. But I frankly cannot imagine that this can cause a problem > of this sort. > > The situation is as follows: > > / > A (nt:unstructured) > A1 (nt:unstructured) > A2 (nt:unstructured) > B (nt:unstructured) > B1 (nt:unstructured) > B2 (nt:unstructured) > > The problem is that no query in the /A substree works anymore. They all > return empty, including statements that have worked for months. Finally, I > tested something like this: > > //element(*, nt:unstructed) > > This query returns all nt:unstructured nodes of B, but not of A! The > behaviour is the same for both, application and query tools. I already > rebuilt the index, but the effect is still the same. > > Question: could there be a Lucene internal lock? I consider that unlikely, > the rebuild would have solved the problem. Or is there an inconcistence in > the JCR that makes a correct indexing impossible? Also unlikely, as the node > tree of A can be navigated without a problem through the JCR API. > > Thus, I'm pretty clouless. Anybody with an idea? > > Regards, > Juerg >
